In the notebook on linear regression, we defined a weights matrix which was 2x3 and then took a transpose of that matrix while multiplying it to input matrix X.

Why didn’t we define a 3x2 matrix itself? Why take transpose instead of doing it the straight way?